Single Hemisphere Segmentation

[akimbler]

I'm running in to an issue and I'd like some advice. We're working with a dataset with multiple post-mortem brains, some have the intact whole brain, while others are limited to a single hemisphere. The segmentation is great for the whole-brain subjects, which makes sense based on the training dataset

It also does a good job segmenting cortical structures, but the rest of the cortex and cerebellum segmentation breaks.

Do you think it would be possible to train the tool to be more robust for such use cases?